Why AI Social Media Automation Doesn't Work (And What To Do Instead)

Key Concepts

• AI Cost & Efficiency: While Gemini 3.1 performs well, its high token usage makes it surprisingly expensive compared to alternatives like Sonnet 4.6. Model “sticker price” is no longer a reliable indicator of actual cost.
• Anthropic’s Enterprise Focus: Anthropic is prioritizing enterprise solutions with Claude and co-work, particularly through the introduction of easily deployable and manageable internal “skills” (plugins).
• AI-Driven Disruption: AI models are rapidly disrupting traditional software industries, demonstrating the potential to replace significant development effort with relatively simple AI implementations.
• The Rise of AI Agents: Platforms like Notion are launching custom agents for automation, but API access from model creators (like Anthropic) remains more cost-effective.
• Automation & the Future of Work: The speed of AI development is accelerating, leading to daily releases and self-improving models, raising concerns about competition and the future of work.

---

Gemini 3.1: Performance vs. Cost

Gemini 3.1, while ranking as the third smartest model currently available, presents a cost challenge. Despite a small price difference on the API between its Opus and Sonnet models (15-20%), Opus consumes significantly more tokens during reasoning, potentially making Sonnet cheaper for complex tasks. This contrasts with previous models like Opus 4.5, which was cheaper to run than Sonnet 4.6 for high-reasoning tasks. Gemini 3.1 is still favored as a default chat model for quick questions and tool usage due to its efficiency and minimal impact on usage limits.

Anthropic’s Enterprise Strategy & Skill Development

Anthropic is heavily investing in enterprise solutions centered around Claude and the co-work platform. A key development is the release of plugins enabling enterprise users to create and deploy internal plugins (“skills”) for their teams. Previously, sharing skills was cumbersome, relying on workarounds like GitHub. The new system offers centralized deployment, version control, and access management. A skill is defined as a collection of text files or Python scripts, with updates seamlessly propagated to all users. This is expected to drive demand for consultancy services focused on skill setup and workflow building.

Co-work vs. Cloud Code

Co-work is positioned as a more accessible alternative to Cloud Code, particularly for users unfamiliar with VS Code. Both platforms allow skills to automatically scale with usage. A real-world example showcased a skill analyzing Salesforce data, generating presentations, initiating legal discussions, and creating action items – demonstrating complete workflow automation. Co-work is predicted to gain significant traction by the end of the year, potentially becoming the dominant platform for knowledge work automation.

AI Disruption of Traditional Software

AI models are rapidly disrupting traditional software industries. A legal skill developed by Claude reportedly caused a $285 billion drop in the stock market value of established legal software companies, demonstrating superior performance with a relatively simple 200-line instruction. This illustrates AI’s potential to replace decades of development effort. The speed of AI development – with daily releases and self-improving models – makes it difficult for traditional companies to compete. This dynamic is likened to Amazon’s strategy of undercutting third-party sellers.

Notion Agents & the Competitive Landscape

Notion has launched custom agents, similar to Open Claude and co-work, allowing users to build automations within Notion using a page of instructions and connecting to tools via MCPs (Master Connector Protocols). However, the cost of running these agents, particularly with models like Opus, is a concern, potentially reaching thousands of dollars per month. This reinforces the argument that direct API access from model creators remains more cost-effective. Notion’s strategy appears geared towards companies with substantial budgets and limited technical expertise. Co-work and the code desktop app are expected to offer comparable automation features, potentially undermining Notion’s competitive advantage.

Future Concerns & Predictions

A central concern is the potential for a race to the bottom in AI development, making it difficult for independent developers to compete. If a startup creates a successful AI product, larger companies with greater resources can quickly replicate it. The discussion concludes with a prediction that Anthropic is currently winning the enterprise market and that competition is underestimating the importance of knowledge work automation. The potential for self-improving AI loops and the sheer volume of AI-generated content are identified as future challenges, requiring AI-powered curation and filtering mechanisms.

---

Conclusion:

The current AI landscape is characterized by rapid development, shifting cost dynamics, and a growing focus on enterprise applications. While models like Gemini 3.1 offer strong performance, cost-effectiveness remains a critical consideration. Anthropic’s strategic push into the enterprise market with Claude and co-work, particularly through the introduction of easily manageable “skills,” positions them as a key player. The potential for AI to disrupt traditional software industries is significant, and the speed of innovation necessitates a proactive approach to adaptation and competition. The future will likely require AI-powered solutions to manage the increasing volume of AI-generated content and ensure continued value creation.